A model of your processes, and the AI opportunities in them.
We help you map selected business processes with their owners, applications, data, controls and objectives; find where AI creates measurable value; and show what it takes to govern it. You keep the model.
The question we hear
“Everyone has an AI idea. Nobody can tell me which ones pay, which ones we can actually run on our data, and what we'd be signing up to govern.”
“We're about to be asked how our AI is governed, and we don't have a clear picture of where it sits in our processes.”
The meta-model behind it
Behind every mapping is Valment's meta-model: the types of thing an organization is made of and how they relate. Objectives, processes and their owners, applications, data, AI systems, risks, controls and performance; a process has an owner, runs in an application, consumes data, carries objectives, contains controls, and may use AI or be analysed by it. What you see here is a part of it, enough to show the principle. The full meta-model is richer, and your model is the part of it we populate together for your processes.
- A process has an owner, carries objectives and KPIs, and is measured against them.
- A process runs in one or more applications, which hold the data the process consumes.
- A process may use AI, or be analysed by it; the AI is fed by the same data.
- A process carries risks; its controls treat them, and are enforced in the applications.
The meta-model is also the foundation of an ontology we use to feed the AI models that do the analysis. They work with your organization's own structure and vocabulary rather than a generic one, which is what lets them perform at their best.
Two roles of AI
Analyses process models and operational data. Finds anomalies, bottlenecks and exceptions. Identifies patterns and improvement potential. Prioritises opportunities on evidence.
Automates or augments activities. Supports decisions and predictions. Introduces intelligent workflows and agents. Delivers measurable operational outcomes.
Both need governance, and the model shows where.
What we do together
- 01
Agree the scope
The selected processes, usually one to three, and what "better" means for each of them.
- 02
Map each process, with its owners
Activities, owners, applications, data, objectives and KPIs, and the controls already in place.
- 03
Analyse what is there
The documentation and the data that is available, AI-assisted, under our own AI policy and with human validation.
- 04
Find the AI
Where AI improves the process, and where AI becomes part of it.
- 05
Evaluate each opportunity
Value, feasibility, data readiness, risk.
- 06
Prioritise, and state the governance
What each opportunity implies: roles, risk and impact assessment, oversight, information security, lifecycle controls.
- 07
Hand over
The model, the portfolio and the roadmap, and a walk-through with your team.
How it runs
Fixed scope on selected processes, agreed in the first conversation. Weeks, not months.
Workshops with the process owners and the people who run the systems; data analysis where access and data quality allow; process mining and transaction analysis as options, never assumed.
We use our own AI tooling in the analysis. The result is reviewed by a person before you see it.
What you get
- A process architecture and the selected process models
- A structured register of processes, applications, AI systems, risks, controls and responsibilities, built on Valment's meta-model, with the relationships between them made explicit
- An AI opportunity portfolio with value, feasibility and risk for each opportunity
- A roadmap and management recommendations
- Delivered in formats you already use: spreadsheets, diagrams and structured documents. No software to buy. The model is yours to maintain and extend.
What you provide
- The process owners' time
- Access to process documentation and to the data we agree on
- One decision-maker
- A view of what you are trying to achieve
- If you run a quality management system, its process objectives, measures and improvement records are the starting data; the mapping goes faster and the value case starts with a baseline.
A first engagement maps selected processes, not the enterprise.
We don't build the AI applications. We help you specify, prioritise and govern them, and can help you choose who builds.
Where an opportunity needs a formal AI risk and impact assessment or an AI management system, that is the next engagement. Where you want the controls tested independently, that is an assessment or audit, by someone who did not design them.
See it on a process
A fictional purchase-to-pay process, mapped the way we would map yours: what the model looks like once populated, which opportunities appeared, which governance they implied, and how the controls would be evidenced. Illustrative, and labelled as such.
The next step would be a conversation about which process is causing the most friction, or carrying the most AI ambition. We'll come back with a fixed scope.
Map your first process
Tell us which process is causing the most friction, or carrying the most AI ambition. We'll come back with a fixed scope on one to three processes.